In [1]:
import pandas as pd
import numpy as np
from numpy import prod

import matplotlib.pyplot as plt
import seaborn as sns
import plotly.express as px
import wordcloud
import sklearn
from wordcloud import WordCloud, STOPWORDS, ImageColorGenerator
from sklearn.model_selection import train_test_split
from sklearn.linear_model import PassiveAggressiveRegressor,LinearRegression
from sklearn.metrics import mean_squared_error, r2_score


data = pd.read_csv("Instagram Data.csv", encoding = 'latin1')
print(data.head())

   Impressions  From Home  From Hashtags  From Explore  From Other  Saves  \
0         3920       2586           1028           619          56     98   
1         5394       2727           1838          1174          78    194   
2         4021       2085           1188             0         533     41   
3         4528       2700            621           932          73    172   
4         2518       1704            255           279          37     96   

   Comments  Shares  Likes  Profile Visits  Follows  \
0         9       5    162              35        2   
1         7      14    224              48       10   
2        11       1    131              62       12   
3        10       7    213              23        8   
4         5       4    123               8        0   

                                             Caption  \
0  Here are some of the most important data visua...   
1  Here are some of the best data science project...   
2  Learn how to train a machine learning model an...   
3  Here’s how you can write a Python program to d...   
4  Plotting annotations while visualizing your da...   

                                            Hashtags  
0  #finance #money #business #investing #investme...  
1  #healthcare #health #covid #data #datascience ...  
2  #data #datascience #dataanalysis #dataanalytic...  
3  #python #pythonprogramming #pythonprojects #py...  
4  #datavisualization #datascience #data #dataana...

Let's first determine whether or not this dataset has any null values before we get started:¶

In [2]:
data.isnull().sum()
Out[2]:
Impressions       0
From Home         0
From Hashtags     0
From Explore      0
From Other        0
Saves             0
Comments          0
Shares            0
Likes             0
Profile Visits    0
Follows           0
Caption           0
Hashtags          0
dtype: int64
In [ ]:

To fully understand the data type of each column, let's examine each column:¶

In [3]:
data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 119 entries, 0 to 118
Data columns (total 13 columns):
 #   Column          Non-Null Count  Dtype 
---  ------          --------------  ----- 
 0   Impressions     119 non-null    int64 
 1   From Home       119 non-null    int64 
 2   From Hashtags   119 non-null    int64 
 3   From Explore    119 non-null    int64 
 4   From Other      119 non-null    int64 
 5   Saves           119 non-null    int64 
 6   Comments        119 non-null    int64 
 7   Shares          119 non-null    int64 
 8   Likes           119 non-null    int64 
 9   Profile Visits  119 non-null    int64 
 10  Follows         119 non-null    int64 
 11  Caption         119 non-null    object
 12  Hashtags        119 non-null    object
dtypes: int64(11), object(2)
memory usage: 12.2+ KB

Now let’s start with analyzing the reach of my Instagram posts. I will first have a look at the distribution of impressions I have received from home:¶

In [4]:
plt.figure(figsize=(10, 8))
plt.style.use('fivethirtyeight')
plt.title("Distribution of Impressions From Home")
sns.distplot(data['From Home'])
plt.show()

C:\Users\vacha\AppData\Local\Temp\ipykernel_3716\4127021947.py:4: UserWarning: 

`distplot` is a deprecated function and will be removed in seaborn v0.14.0.

Please adapt your code to use either `displot` (a figure-level function with
similar flexibility) or `histplot` (an axes-level function for histograms).

For a guide to updating your code to use the new functions, please see
https://gist.github.com/mwaskom/de44147ed2974457ad6372750bbe5751

  sns.distplot(data['From Home'])
No description has been provided for this image

The impressions I get from the home section on Instagram shows how much my posts reach my followers. Looking at the impressions from home, I can say it’s hard to reach all my followers daily. Now let’s have a look at the distribution of the impressions I received from hashtags:¶

In [5]:
plt.figure(figsize=(10, 8))
plt.title("Distribution of Impressions From Hashtags")
sns.distplot(data['From Hashtags'])
plt.show()

C:\Users\vacha\AppData\Local\Temp\ipykernel_3716\671336047.py:3: UserWarning: 

`distplot` is a deprecated function and will be removed in seaborn v0.14.0.

Please adapt your code to use either `displot` (a figure-level function with
similar flexibility) or `histplot` (an axes-level function for histograms).

For a guide to updating your code to use the new functions, please see
https://gist.github.com/mwaskom/de44147ed2974457ad6372750bbe5751

  sns.distplot(data['From Hashtags'])
No description has been provided for this image

Hashtags are tools we use to categorize our posts on Instagram so that we can reach more people based on the kind of content we are creating. Looking at hashtag impressions shows that not all posts can be reached using hashtags, but many new users can be reached from hashtags. Now let’s have a look at the distribution of impressions I have received from the explore section of Instagram:¶

In [6]:
plt.figure(figsize=(10, 8))
plt.title("Distribution of Impressions From Explore")
sns.distplot(data['From Explore'])
plt.show()

C:\Users\vacha\AppData\Local\Temp\ipykernel_3716\2610232579.py:3: UserWarning: 

`distplot` is a deprecated function and will be removed in seaborn v0.14.0.

Please adapt your code to use either `displot` (a figure-level function with
similar flexibility) or `histplot` (an axes-level function for histograms).

For a guide to updating your code to use the new functions, please see
https://gist.github.com/mwaskom/de44147ed2974457ad6372750bbe5751

  sns.distplot(data['From Explore'])
No description has been provided for this image

The explore section of Instagram is the recommendation system of Instagram. It recommends posts to the users based on their preferences and interests. By looking at the impressions I have received from the explore section, I can say that Instagram does not recommend our posts much to the users. Some posts have received a good reach from the explore section, but it’s still very low compared to the reach I receive from hashtags.¶

Now let’s have a look at the percentage of impressions I get from various sources on Instagram:¶

In [7]:
home = data["From Home"].sum()
hashtags = data["From Hashtags"].sum()
explore = data["From Explore"].sum()
other = data["From Other"].sum()

labels = ['From Home','From Hashtags','From Explore','Other']
values = [home, hashtags, explore, other]

fig = px.pie(data, values=values, names=labels, 
             title='Impressions on Instagram Posts From Various Sources', hole=0.5)
fig.show()

So the above donut plot shows that almost 44.1 per cent of the reach is from my followers, 33.6 per cent is from hashtags, 19.2 per cent is from the explore section, and 3.05 per cent is from other sources.¶

Now let’s analyze the content of my Instagram posts. The dataset has two columns, namely caption and hashtags, which will help us understand the kind of content I post on Instagram.¶

Let’s create a wordcloud of the caption column to look at the most used words in the caption of my Instagram posts:¶

In [8]:
text = " ".join(i for i in data.Caption)
stopwords = set(STOPWORDS)
wordcloud = WordCloud(stopwords=stopwords, background_color="white").generate(text)
plt.style.use('classic')
plt.figure( figsize=(12,10))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()
No description has been provided for this image

Now let’s create a wordcloud of the hashtags column to look at the most used hashtags in my Instagram posts:¶

In [9]:
text = " ".join(i for i in data.Hashtags)
stopwords = set(STOPWORDS)
wordcloud = WordCloud(stopwords=stopwords, background_color="white").generate(text)
plt.figure( figsize=(12,10))
plt.imshow(wordcloud, interpolation='bilinear')
plt.axis("off")
plt.show()
No description has been provided for this image

Now let’s analyze relationships to find the most important factors of our Instagram reach. It will also help us in understanding how the Instagram algorithm works.¶

Let’s have a look at the relationship between the number of likes and the number of impressions on my Instagram posts:¶

In [10]:
figure = px.scatter(data_frame = data, x="Impressions",
                    y="Likes", size="Likes", trendline="ols", 
                    title = "Relationship Between Likes and Impressions")
figure.show()

There is a linear relationship between the number of likes and the reach I got on Instagram. Now let’s see the relationship between the number of comments and the number of impressions on my Instagram posts:¶

In [11]:
figure = px.scatter(data_frame = data, x="Impressions",
                    y="Comments", size="Comments", trendline="ols", 
                    title = "Relationship Between Comments and Total Impressions")
figure.show()

It looks like the number of comments we get on a post doesn’t affect its reach. Now let’s have a look at the relationship between the number of shares and the number of impressions:¶

In [12]:
figure = px.scatter(data_frame = data, x="Impressions",
                    y="Shares", size="Shares", trendline="ols", 
                    title = "Relationship Between Shares and Total Impressions")
figure.show()

A more number of shares will result in a higher reach, but shares don’t affect the reach of a post as much as likes do. Now let’s have a look at the relationship between the number of saves and the number of impressions:¶

In [13]:
figure = px.scatter(data_frame = data, x="Impressions",
                    y="Saves", size="Saves", trendline="ols", 
                    title = "Relationship Between Post Saves and Total Impressions")
figure.show()

There is a linear relationship between the number of times my post is saved and the reach of my Instagram post. Now let’s have a look at the correlation of all the columns with the Impressions column:¶

In [14]:
correlation = data.corr()
print(correlation["Impressions"].sort_values(ascending=False))

Impressions       1.000000
From Explore      0.893607
Follows           0.889363
Likes             0.849835
From Home         0.844698
Saves             0.779231
Profile Visits    0.760981
Shares            0.634675
From Other        0.592960
From Hashtags     0.560760
Comments         -0.028524
Name: Impressions, dtype: float64

C:\Users\vacha\AppData\Local\Temp\ipykernel_3716\3935629544.py:1: FutureWarning:

The default value of numeric_only in DataFrame.corr is deprecated. In a future version, it will default to False. Select only valid columns or specify the value of numeric_only to silence this warning.

So we can say that more likes and saves will help you get more reach on Instagram. The higher number of shares will also help you get more reach, but a low number of shares will not affect your reach either¶

In Instagram, conversation rate means how many followers you are getting from the number of profile visits from a post. The formula that you can use to calculate conversion rate is (Follows/Profile Visits) * 100. Now let’s have a look at the conversation rate of my Instagram account:¶

In [15]:
conversion_rate = (data["Follows"].sum() / data["Profile Visits"].sum()) * 100
print(conversion_rate)

41.00265604249668

So the conversation rate of my Instagram account is 41% which sounds like a very good conversation rate. Let’s have a look at the relationship between the total profile visits and the number of followers gained from all profile visits:¶

In [16]:
figure = px.scatter(data_frame = data, x="Profile Visits",
                    y="Follows", size="Follows", trendline="ols", 
                    title = "Relationship Between Profile Visits and Followers Gained")
figure.show()

The relationship between profile visits and followers gained is also linear.¶

Instagram Reach Prediction Model¶

Now in this section, I will train a machine learning model to predict the reach of an Instagram post. Let’s split the data into training and test sets before training the model:¶

In [17]:
x = np.array(data[['Likes', 'Saves', 'Comments', 'Shares', 
                   'Profile Visits', 'Follows']])
y = np.array(data["Impressions"])
xtrain, xtest, ytrain, ytest = train_test_split(x, y, 
                                                test_size=0.2, 
                                                random_state=42)

Now here’s is how we can train a machine learning model to predict the reach of an Instagram post using Python:¶

In [18]:
model = LinearRegression()
model.fit(xtrain, ytrain)
y_pred = model.predict(xtest)

Now let’s predict the reach of an Instagram post by giving inputs to the machine learning model:¶

In [19]:
# Features = [['Likes','Saves', 'Comments', 'Shares', 'Profile Visits', 'Follows']]
features = np.array([[282.0, 233.0, 4.0, 9.0, 165.0, 54.0]])
model.predict(features)
Out[19]:
array([11139.58239766])

Here is the metrics for the model¶

In [20]:
mse = mean_squared_error(ytest, y_pred)
r2 = r2_score(ytest, y_pred)

print(f"Mean Squared Error: {mse}")
print(f"R-squared: {r2}")

Mean Squared Error: 4746059.117340458
R-squared: 0.8777977785012778
In [ ]:
In [ ]:
In [ ]: